Electron diffraction tomography (EDT) data are in many ways similar to X-ray diffraction data. However, they also present certain specifics. One of the most noteworthy is the specific rocking curve observed for EDT data collected using the precession electron diffraction method. This double-peaked curve (dubbed 'the camel') may be described with an approximation based on a circular integral of a pseudo-Voigt function and used for intensity extraction by profile fitting. Another specific aspect of electron diffraction data is the high likelihood of errors in the estimation of the crystal orientation, which may arise from the inaccuracies of the goniometer reading, crystal deformations or crystal movement during the data collection. A method for the refinement of crystal orientation for each frame individually is proposed based on the least-squares optimization of simulated diffraction patterns. This method provides typical angular accuracy of the frame orientations of less than 0.05 . These features were implemented in the computer program PETS 2.0. The implementation of the complete data processing workflow in the program PETS and the incorporation of the features specific for electron diffraction data is also described. electron crystallography 514 L. Palatinus et al. Structure refinement from PED Acta Cryst. (2019). B75, 512-522 electron crystallography Acta Cryst. (2019). B75, 512-522 L. Palatinus et al. Structure refinement from PED 517 electron crystallography Acta Cryst. (2019). B75, 512-522 L. Palatinus et al. Structure refinement from PED 521

The recently published method for the structure refinement from threedimensional precession electron diffraction data using dynamical diffraction theory [Palatinus et al. (2015). Acta Cryst. A71, 235-244] has been applied to a set of experimental data sets from five different samples -Ni 2 Si, PrVO 3 , kaolinite, orthopyroxene and mayenite. The data were measured on different instruments and with variable precession angles. For each sample a reliable reference structure was available. A large series of tests revealed that the method provides structure models with an average error in atomic positions typically between 0.01 and 0.02 Å . The obtained structure models are significantly more accurate than models obtained by refinement using kinematical approximation for the calculation of model intensities. The method also allows a reliable determination of site occupancies and determination of absolute structure. Based on the extensive tests, an optimal set of the parameters for the method is proposed.

Transition metal oxides having a perovskite structure form a wide and technologically important class of compounds. In these systems, ferroelectric, ferromagnetic, ferroelastic, or even orbital and charge orderings can develop and eventually coexist. These orderings can be tuned by external electric, magnetic, or stress field, and the cross-couplings between them enable important multifunctional properties, such as piezoelectricity, magneto-electricity, or magneto-elasticity. Recently, it has been proposed that additional to typical fields, the chemical potential that controls the concentration of ion vacancies in these systems may reveal an efficient alternative parameter to further tune their properties and achieve new functionalities. In this study, concretizing this proposal, the authors show that the control of the content of oxygen vacancies in perovskite thin films can indeed be used to tune their magnetic properties. Growing PrVO thin films epitaxially on an SrTiO substrate, the authors reveal a concrete pathway to achieve this effect. The authors demonstrate that monitoring the concentration of oxygen vacancies through the oxygen partial pressure or the growth temperature can produce a substantial macroscopic tensile strain of a few percent. In turn, this strain affects the exchange interactions, producing a nontrivial evolution of Néel temperature in a range of 30 K.